When applying wavelet transformation method in the retrieval of the boundary layer height by using lidar backscatter signals, the different selection of wavelet generating function may get different results.Therefore,the idealized lidar signal profiles during the daytime and nighttime have been built to explore which wavelet generating function will get the best performance.In this study,Haar wavelet covariance transformation has been used for the profiles of lidar backscatter signal,and Morlet and Mexican Hat wavelet transformation have been utilized for the gradient profiles of lidar backscatter signal,respectively.The results showed that the Haar function and the Mexican Hat function should be used as wavelet generating functions:the Haar function is more accurate and the Mexican Hat function is more stable. Furthermore,the changed wavelet dilation of the wavelet generating function also has been researched in order to test the sensitivity of the three different wavelet transformation methods to the wavelet dilation. The result show that whatever the idealized profile or the profile in which disturbance was added,the larger wavelet dilation can get a more stable and accurate diurnal boundary layer height and nocturnal mixing layer height.
The simulanted radar data are produced by using high-order Lengendre polynomialst of it the real radar observed wind. Three interpolation schemes are tested with the simulated data.In first scheme(CVI),the radar data are first interpolated to horizontal Cartesian grids but left on the iroriginal constant elevation surfaces in the vertical,then the data are further interpolated to the Cartesian grids with a linear interpolation in the vertical. The second schemeis the three-dimension Barnes interpolation technique(3DBarnes).The third scheme is three-dimension variational analysis(VAR).The results show that the3D-Barnes scheme obviously excels the others,especially at middle level;the error of the CVI scheme is slightly larger than the former,and its smooth nessispoor, and it has biggish missing region at lower and upper level.The VAR scheme is not perfect.